{"record": {"aliases": [], "artifacts": [{"id": "arxiv:2605.30058", "kind": "paper", "url": "https://arxiv.org/abs/2605.30058"}, {"id": "gh:peng-weihan/heart-bench", "kind": "repo", "url": "https://github.com/peng-weihan/HEART-BENCH"}], "authors": ["Weihan Peng", "Chenxu Zhang", "Qianao Wang", "Yuling Shi", "Heng Lian", "Qihong Mao", "Jiahao Pang", "Chunliang Feng", "Bowen Li", "Xiaodong Gu"], "authors_reference": {"basis": "arxiv_metadata", "retrieved_at": "2026-11-03", "source_key": "claire-radar:2605.30058", "source_url": "https://arxiv.org/abs/2605.30058"}, "categories": ["General AI", "Language & Knowledge", "cs.CL"], "description": {"en": "HEART-Bench evaluates whether LLM agents can simulate coherent, human-like psychology. It provides 11 fictional characters with raw episodic memories, 64 decision-making scenarios based on the DIAMONDS taxonomy, and 673 multiple-choice questions with expert-annotated ground truth. Two evaluation tracks: MCQ and open-ended consciousness-narrative."}, "documents": [{"document_type": "registry_page", "id": "claire_radar:https://arxiv.org/abs/2605.30058", "source": "claire_radar", "source_url": "https://arxiv.org/abs/2605.30058", "title": "HEART-Bench"}], "evidence_summary": {"document_count": 1, "model_count": null, "model_count_basis": null}, "institutions": ["Shanghai Jiao Tong University", "Imperial College London", "University of Washington", "South China Normal University"], "institutions_reference": {"basis": "openalex_linked_institutions", "doi": "10.48550/arxiv.2605.30058", "retrieved_at": "2026-11-04", "source_key": "claire-radar:2605.30058", "source_url": "https://openalex.org/W7162802760"}, "key": "claire-radar:2605.30058", "modality": null, "name": "HEART-Bench", "openness": {"code_license": null, "data_license": null, "evidence": [], "status": "unknown"}, "provenance": {"confidence": "High", "crawl_bundle": "claire_radar_2026-10-14", "crawled_at": "2026-10-14T09:44:51Z", "data_status": "primary-source-indexed", "display_eligible": "unknown", "origin_record_id": "bm_heart-bench_230e27fe", "origin_source": "arxiv", "recognition_confidence": "1.0", "record_sha256": "4887a83e835ce6e6693d78ca8db4b7df429a7f48b14ec5e2cfdf90aa9fa7de8b", "relation": "introduces", "review_state": "unreviewed", "source_url": "https://arxiv.org/abs/2605.30058"}, "publisher": null, "released": "2026-05-28", "released_reference": {"basis": "benchmark_release", "source_key": "claire-radar:2605.30058", "source_url": "https://arxiv.org/abs/2605.30058"}, "schema_version": 1, "sizes": [], "slug": "claire-radar-2605-30058", "source": "claire_radar", "source_benchmark_id": "2605.30058", "source_metadata": {"claire_radar": {"adoption30d": null, "aliases": [], "annotation": "Unknown", "applicationDomains": ["General AI"], "area": "Language & Knowledge", "attention": {"asOf": "2026-10-05", "attemptedAt": "2026-10-04T15:31:48.030916Z", "githubRepo": "https://github.com/peng-weihan/HEART-BENCH", "githubScope": "benchmark_repo", "githubStars": 4, "hfDailySubmittedAt": null, "hfDataset": null, "hfDatasetDownloads": null, "hfDatasetLikes": null, "hfPaperUpvotes": null, "hfPaperUrl": "https://huggingface.co/papers/2605.30058", "observedAt": "2026-10-04T15:31:48.030916Z", "signalStatus": {"githubStars": {"observedAt": "2026-10-04T15:31:48.030916Z", "state": "fresh"}, "hfDatasetDownloads": {"state": "not_applicable"}, "hfDatasetLikes": {"state": "not_applicable"}, "hfPaperUpvotes": {"checkedAt": "2026-10-04T15:31:48.030916Z", "state": "unavailable"}}, "status": "fresh"}, "capabilities": [], "confidence": "High", "construction": "Transform Existing", "constructionDetail": "Unknown — the indexer does not infer construction details without explicit source evidence.", "copyGeneration": {"generatedAt": "2026-08-20T15:22:10.044731Z", "inputHash": "60ecad22cf0bf9918ea5ec0cf1b21cea1621810990694d503c3a0440560ee793", "model": "deepseek-v4-flash"}, "dataStatus": "primary-source-indexed", "demo": false, "description": "HEART-Bench evaluates whether LLM agents can simulate coherent, human-like psychology. It provides 11 fictional characters with raw episodic memories, 64 decision-making scenarios based on the DIAMONDS taxonomy, and 673 multiple-choice questions with expert-annotated ground truth. Two evaluation tracks: MCQ and open-ended consciousness-narrative.", "domainCuration": {"method": "rules-v1", "state": "auto"}, "evidence": {"reasonCodes": ["coined title prefix ending in Bench or Benchmark", "benchmark release stated in one sentence", "evaluation protocol evidence", "public artifact URL"], "snippet": "In this paper, we introduce a novel benchmark to systematically assess whether LLM agents can simulate coherent, human-like psychology."}, "familyId": "bmf_1835f35879b7", "firstSeenAt": "2026-08-19", "githubIndex": {"candidateCount": 0, "indexedAt": "2026-09-17T04:40:34.111665+00:00", "repositories": [{"evidence": [{"basis": "existing-code-link", "linkedUrl": "https://github.com/peng-weihan/HEART-BENCH", "url": "https://github.com/peng-weihan/HEART-BENCH"}, {"basis": "existing-metric-association", "linkedUrl": "https://github.com/peng-weihan/HEART-BENCH", "url": "https://huggingface.co/papers/2605.30058"}, {"basis": "linked-from-report", "linkedUrl": "https://github.com/peng-weihan/HEART-BENCH", "url": "https://arxiv.org/abs/2605.30058"}], "scope": "benchmark_repo", "url": "https://github.com/peng-weihan/HEART-BENCH", "verification": "source-linked"}], "status": "linked"}, "heat": null, "id": "bm_heart-bench_230e27fe", "indexedAt": "2026-08-19T11:05:13.395391Z", "industrySectors": [], "links": {"code": "https://github.com/peng-weihan/HEART-BENCH", "data": null, "hfPaper": "https://huggingface.co/papers/2605.30058", "paper": "https://arxiv.org/abs/2605.30058", "pdf": "https://arxiv.org/pdf/2605.30058", "project": null, "report": "https://arxiv.org/abs/2605.30058"}, "metrics": [], "motivation": "While LLM agents have demonstrated remarkable task-oriented abilities such as planning, reasoning, and action, few works have treated them as complete human personalities where emotional dimensions hold equal importance.", "name": "HEART-Bench", "oneLine": "HEART-Bench evaluates whether LLM agents can simulate coherent, human-like psychology. It provides 11 fictional characters with raw episodic memories, 64 decision-making scenarios based on the DIAMONDS taxonomy, and 673 multiple-choice questions with expert-annotated ground truth. Two evaluation tracks: MCQ and open-ended consciousness-narrative.", "paperTitle": "HEART-Bench: Do LLM Agents Exhibit Human-like Psychology?", "primaryDomain": "General AI", "publication": {"evidence": null, "evidenceLevel": "unverified", "evidenceUrl": "https://arxiv.org/abs/2605.30058", "source": "arxiv-metadata", "status": "unverified", "venue": null, "verifiedAt": "2026-10-14T01:46:58.058199Z"}, "publications": [], "publishers": [{"name": "HEART-Bench team", "organizationType": "academic-lab", "role": "benchmark-publisher", "sourceUrl": "https://github.com/peng-weihan/HEART-BENCH"}], "ranking": {}, "readiness": "Runnable", "recognitionConfidence": 1.0, "relation": "introduces", "releasedAt": "2026-05-28", "source": {"authors": ["Weihan Peng", "Chenxu Zhang", "Qianao Wang", "Yuling Shi", "Heng Lian", "Qihong Mao", "Jiahao Pang", "Chunliang Feng", "Bowen Li", "Xiaodong Gu"], "categories": ["cs.CL"], "id": "2605.30058", "title": "HEART-Bench: Do LLM Agents Exhibit Human-like Psychology?", "type": "arxiv", "url": "https://arxiv.org/abs/2605.30058"}, "sourceUpdatedAt": "2026-05-28T15:08:03Z", "topics": ["cs.CL"], "venueAttempts": [], "whyItMatters": "This benchmark addresses the gap in evaluating emotional and personality consistency in LLM agents, complementing task-oriented benchmarks. It offers a standardized protocol for assessing psychological coherence, which is valuable for developing agents that can maintain stable personas and make value-consistent decisions in interactive applications."}}}, "schema_version": 1, "scores_by_source": {}, "siblings": []}
